Is the Indirect Detection of Extrasolar Water Possible? Melissa A. Morris1,2 and Steven J. ..... correspond to the location of the Solar System's asteroid belt.
ASTROBIOLOGY Volume 9, Number 10, 2009 ª Mary Ann Liebert, Inc. DOI: 10.1089=ast.2008.0316
Phyllosilicate Emission from Protoplanetary Disks: Is the Indirect Detection of Extrasolar Water Possible? Melissa A. Morris1,2 and Steven J. Desch1
Abstract
Phyllosilicates are hydrous minerals formed by interaction between rock and liquid water, and are commonly found in meteorites that originate in the asteroid belt. Collisions between asteroids contribute to zodiacal dust, which therefore reasonably could include phyllosilicates. Collisions between planetesimals in protoplanetary disks may also produce dust that contains phyllosilicates. These minerals possess characteristic emission features in the mid-infrared and could be detectable in extrasolar protoplanetary disks. We have determined whether phyllosilicates in protoplanetary disks are detectable in the infrared, using instruments such as those on board the Spitzer Space Telescope and the Stratospheric Observatory for Infrared Astronomy (SOFIA). We calculated opacities for the phyllosilicates most common in meteorites and, using a two-layer radiative transfer model, computed the emission of radiation from a protoplanetary disk. We found that phyllosilicates present at the 3% level lead to observationally significant differences in disk spectra and should therefore be detectable with the use of infrared observations and spectral modeling. Detection of phyllosilicates in a protoplanetary disk would be diagnostic of liquid water in planetesimals in that disk and would demonstrate similarity to our own Solar System. We also discuss use of phyllosilicate emission to test the ‘‘water worlds’’ hypothesis, which proposes that liquid water in planetesimals should correlate with the inventory of short-lived radionuclides in planetary systems, especially 26Al. Key Words: Water—Protoplanetary disks—Infrared emission—Phyllosilicates. Astrobiology 9, 965–978.
Introduction Terrestrial planets and water
I
n the search for extraterrestrial life, it is extremely important to attempt to detect water in extrasolar planetary systems, particularly on terrestrial planets. The central requirements for life as we know it are a source of free energy, a source of carbon, and liquid water (Chyba et al., 2000). Chyba et al. (2000) purported that ‘‘where there is liquid water, there is the possibility of life as we know it.’’ So far, the detection of water on a terrestrial exoplanet has not been achieved, though there are hints that liquid water may occur on such planets. Close-in extrasolar giant planets (hot Jupiters) have been observed to have water vapor in their atmospheres (Beaulieu et al., 2008), and water vapor emission from a protoplanetary disk was recently observed as well (Watson et al., 2007; Carr and Najita, 2008). However, the search for liquid water on Earth-like planets is extremely difficult. Should an Earth-like planet be discovered, strategies exist for detecting liquid water on the surface (Williams and Gaidos, 2008). To date, no
such planets have been found; the most Earth-like planet yet discovered is Gliese 581c (Selsis et al., 2007; Udry et al., 2007), which is nearly 5 times the mass of Earth. Another approach to the search for liquid water on terrestrial planets is to follow the path upstream to the water’s source. While not universally accepted, it is generally thought that the majority of Earth’s water was delivered by planetesimals from the outer asteroid belt (Morbidelli et al., 2000; Raymond et al., 2004; Mottl et al., 2007). We refer the reader to these papers, whose arguments we attempt to summarize here. Comets have long been suggested as a source of Earth’s water (Owen and Bar-Nun, 1995), but the low terrestrial D=H ratio suggests otherwise (Drake and Righter, 2002). In fact, in the few comets for which the D=H ratio has been measured, it is twice that for Earth; the D=H ratio would not be reduced by chemical fractionation processes (Eberhardt et al., 1995; Bockelee-Morvan et al., 1998; Meier et al., 1998). In addition, comets are predicted to introduce too much Ar and other noble gases to be consistent with the low terrestrial Ar=H2O ratio (Owen and Bar-Nun,
1
School of Earth and Space Exploration, Arizona State University, Tempe, Arizona. Present address: Department of Physics, Astronomy, and Materials Science, Missouri State University, Springfield, Missouri.
2
965
966
MORRIS AND DESCH
1995; Swindle and Kring, 1997; Morbidelli et al., 2000). Finally, the likelihood of sufficient comets colliding with Earth is too low to account for Earth’s volatile content (Morbidelli et al., 2000; Levison et al., 2001). Drake and Righter (2002) claimed that, during its formation, Earth received no more than 50% of its water from comets; based on dynamical arguments, Morbidelli et al. (2000) claimed the fraction is less than 10% of Earth’s present water budget. Both Drake and Righter (2002) and Morbidelli et al. (2000) instead strongly argued that the Earth accreted ‘‘wet,’’ with asteroids or planetesimals from the outer asteroid belt as the main source of the water. The D=H ratio in Vienna standard mean ocean water is consistent with that of carbonaceous chondrites, both having D=H ratios of 150 ppm (Drake and Righter, 2002). As reviewed by Morbidelli et al. (2000), carbonaceous chondrites, which are associated spectrally with C-type asteroids (Gradie and Tedesco, 1982) and believed to have formed in the outer asteroid belt (i.e., beyond 2.5 astronomical units), contain *10 wt % water (structurally bound in clays), whereas ordinary and enstatite chondrites, which are associated spectrally with S- and E-type asteroids from the inner belt (Gradie and Tedesco, 1982), contain *0.5–0.1 wt % water. Accretion of a few percent of Earth’s mass from carbonaceous chondritic material from beyond 2.5 astronomical units (AU) is sufficient to explain Earth’s volatile content (Morbidelli et al., 2000; Mottl et al., 2007). Complications to this hypothesis include the difference in oxygen isotopic content between Earth and carbonaceous chondrites and the abundances of siderophiles like Os carried by carbonaceous chondrites (Drake and Righter, 2002). However, reasonable refutations to these objections exist (Mottl et al., 2007). Accordingly, we have assumed that Earth and extrasolar terrestrial planets acquired their water during accretion, from planetesimals. If this is the case, then the volatile content of the planetesimals themselves largely governs how much water or other volatiles a planet will eventually possess. Of course, volatiles may be lost during impacts, and the fraction of water sequestered in the mantle versus that which is outgassed to the surface is not known. All other things being equal, if the planetesimals that make up a terrestrial planet have twice the amount of water than the planetesimals that made up Earth, it may reasonably be expected that such a planet would have twice as much water in its oceans than does Earth. Planetesimal volatiles and
26
Al
Desch and Leshin (2004) pointed out that, if the volatile content of terrestrial planets is determined by the volatile content of the planetesimals from which they formed, then ultimately it will depend on the abundance of 26Al in the planetary system. A general consensus is that the internal heating of asteroids in the Solar System was due to the presence of radioactive 26Al (t1=2 ¼ 0.7 Myr) in these bodies (Grimm and McSween, 1993; Lugmair and Shukolyukov, 2001; Huss et al., 2001; McSween et al., 2002; Gilmour and Middleton, 2009). Grimm and McSween (1993) explained the heliocentric zoning of the asteroid belt based on this heat source. At the time of the formation of calcium-rich, aluminum-rich inclusions (CAIs), the 26Al abundance in the Solar System was 26Al=27Al ¼ 5105 (MacPherson et al., 1995). Asteroids that reached sizes that led to the trapping of
radiogenic heat (roughly 30 km in diameter) in the first 2.6 Myr after CAIs formed would thus have incorporated enough live 26Al (26Al=27Al > 4106) to differentiate completely, as was likely the case for Vesta. Asteroids that grew this large from 2.6–4.5 Myr after the formation of CAIs at the beginning of the Solar System would have incorporated less (4106 > 26Al=27Al > 5107) live 26Al, enough to melt water ice but not rock. Such asteroids would be abundant in products of aqueous alteration, in particular, phyllosilicates. If peak temperatures in the asteroid exceeded 4008C, these phyllosilicates would have decomposed and devolatilized. Such asteroids would resemble S-type asteroids, the presumed parent bodies of ordinary chondrites. If peak temperatures did not exceed 4008C, asteroids within which water ice melted would have retained phyllosilicates and would resemble C-type asteroids, the presumed parent bodies of carbonaceous chondrites. Support for heating as the cause of devolatilization comes from the observed dichotomy inherent in the van Schmus-Wood classification scheme of chondrites; that they either retained abundant hydrated phases (petrologic types 1 and 2) or were heated above 4008C (petrologic types 3–6) (van Schmus and Wood, 1967; Weisberg et al., 2006). Petrologic evidence from CV chondrites strongly indicates they devolatilized by heating (Krot et al., 1995; Kojima and Tomeoka, 1996). In asteroids that incorporated even less live 26Al (26Al=27Al < 5107), ice would never have melted, and phyllosilicates would not have been produced. These asteroids presumably would resemble the D- and P-type asteroids of the outer asteroid belt. Radiogenic heating by 26Al is a major control on planetesimal volatile inventory (Grimm and McSween, 1993; see also Gilmour and Middleton, 2009). Grimm and McSween (1993) argued that the heliocentric zoning of the asteroid belt is a result of varying amounts of incorporated live 26Al, due to increasing accretion times with increasing distance from the Sun. Asteroids inside 2.7 AU grew quickly enough (700 K, are little affected by the Sun’s luminosity). Phyllosilicates apparently are produced on asteroids with just enough 26Al to melt ice but not so much 26Al that the asteroids devolatilize: 5107 < 26Al=27Al < 4106. In our Solar System, this constraint implies that the 26Al in CAIs decayed for 3–5 Myr from its initial value of 26Al=27Al ¼ 5105. It is highly likely that this initial abundance of 26Al was fixed by the amount of material the Solar System incorporated from a nearby supernova (Lee et al., 1976; Hester et al., 2004; Jacobsen, 2005; Wadhwa et al., 2007), either from ejecta that contaminated its molecular cloud core (Cameron and Truran, 1977; Boss and Vanhala, 2000) or from ejecta injected into its already formed protoplanetary disk (Ouellette et al., 2005, 2007; see also Looney et al., 2006). Either way, injection of supernova material into a forming planetary
PHYLLOSILICATE EMISSION system is a highly stochastic process, and other planetary systems are likely to have very different initial 26Al abundances. Specifically, protoplanetary disks in regions that lack a massive star (e.g., the Taurus-Auriga region) will incorporate no 26Al from a nearby supernova and almost certainly have 26Al=27Al ratios orders of magnitude lower than was the case in our Solar System. In a system with 26Al=27Al < 5107, ice would not melt on any planetesimals. Any planetesimals that formed in a part of a protoplanetary disk where ice condensed would contain abundant water ice and resemble D- and P-type asteroids. (The disk temperatures in a typical disk will not exceed the sublimation temperature of ice, 180 K, outside of about 0.7 AU (Chiang and Goldreich, 1997.) In such systems, no phyllosilicates would be produced on any planetesimals. On the other hand, terrestrial planets that form in such systems, mostly from planetesimals inside 2.6 AU (Raymond et al., 2004, 2006), are likely to contain substantially more water than Earth, perhaps closer to tens of percent by weight. Planets in such 26Al-poor systems would be ‘‘water worlds,’’ the internal structures of which were explored by Le´ger et al. (2004). 26Al-rich systems, like our Solar System, would possess planetesimals where ice melted and phyllosilicates were produced. Terrestrial planets in such systems are predicted to be ‘‘dry,’’ like Earth. Likely regions to search for phyllosilicate emission from such systems would be protoplanetary disks in the Orion Ic and Id subgroups, whose O and Si abundances strongly suggest contamination by supernovae in the Orion Ia and Ib subgroups (Cunha and Lambert, 1992, 1994; Cunha et al., 1998). The hypothesis that planetesimals in star-forming regions like Taurus will be ice-rich and result in waterrich terrestrial planets, or ‘‘water worlds,’’ admittedly is based on a number of assumptions. One prediction of this water worlds hypothesis that can be tested is the expectation that the dust in protoplanetary disks in regions like Taurus-Auriga will not exhibit phyllosilicates. We predict that phyllosilicates are possible only in systems that formed near a supernova, with abundant 26Al. If the material in 26 Al-rich, phyllosilicate-bearing planetesimals is returned to the protoplanetary disk, and if phyllosilicates can be spectrally distinguished, then phyllosilicate emission is predicted to be found in these 26Al-rich systems, and in these systems only. Mid-infrared (MIR) spectra Strong observational evidence for disks around young stellar objects (YSOs) and T Tauri stars exists, especially in the form of excess infrared emission over what would be expected from the stellar photosphere alone (Adams et al., 1988; McCaughrean and O’Dell, 1996; Chiang and Goldreich, 1997; de Pater and Lissauer, 2001). In fact, IR excess emission is observed in 25–50% of pre-main-sequence stars of 1 solar mass (M) (de Pater and Lissauer, 2001). The excess IR emission is the result of thermal emission due to reprocessed starlight from circumstellar dust grains (Adams et al., 1988; Hartmann, 1998). During the T Tauri stage, the system consists of a passive, reprocessing disk, in which the excess infrared emission originates from dust grains in the outer layers of the disk that are heated by starlight (Kenyon and Hartmann, 1987).
967 The spectra of YSOs with IR excesses contain information about the composition of the dust that gives rise to that emission. While observations at millimeter wavelengths probe closer to the midplane of the disk, MIR observations probe the surface layers of the disk. The spectra of YSOs routinely exhibit silicate emission bands at l * 10 mm and l * 20 mm. This implies that the grains are emitting in the Rayleigh limit, a l=2p, where a is the radius of the grain, as the observed silicate feature should diminish if the particle was more than a few microns in size (Pollack et al., 1994; Nakamura, 1998). The silicate band positions and profiles are highly diagnostic of stoichiometry (Dorschner et al., 1995; Fabian et al., 2001; Krishna Swamy, 2005). Phyllosilicates exhibit the 10 and 20 mm features characteristic of silicates, with distinctive substructure particular to each specific mineral (Fig. 1). All phyllosilicates also show a distinct absorption feature at 6 mm and other wavelengths due to structural H2O. The unique and distinctive MIR spectral features of silicates in general and phyllosilicates in particular make the study of the mineralogy of protoplanetary disks possible. The disk environment Forming planetary systems are generally observed in one of two states: the protoplanetary disk stage, when gas and dust are both present and presumed similar in properties to interstellar material; and the debris disk stage, after gas has been removed and only dust from collisions between planetesimals is present. Examples of protoplanetary disks are abundant (Adams et al., 1987) and include the archetype T Tauri. The fraction of disks in a cluster observed to possess protoplanetary disks tends to decrease with age of the cluster, dropping below 50% at 3–6 Myr (Haisch et al., 2001). Examples of debris disks include b Pic (Smith and Terrile, 1984) and AU Mic (Kalas et al., 2004), both of which are members of the 12 Myr old b Pic moving group (Zuckerman et al., 2001). Objects apparently in transition between the two also exist, such as TW Hya (Calvet et al., 2002). Since the formation of terrestrial planets takes several tens of Myr (Wadhwa and Russell, 2000), debris disks would, ideally, be better to observe, as they would contain only dust from planetesimals at the time terrestrial planets are forming. Unfortunately, debris disks are too faint to search for the infrared signatures of phyllosilicates. The average column density of debris disks ranges from *104 to *107 g cm2, as compared to the column density in the superheated dust layer of a protoplanetary disk, which is on the order of *102 g cm2 (Chiang and Goldreich, 1997). The dust emission features (which can arise only in the optically thin portions of disks) are therefore several orders of magnitude weaker in debris disks than in protoplanetary disks. The fluxes from the nearest debris disk, b Pictoris, are sufficiently large to detect the dominant emission features, such as crystalline and amorphous silicates (Okamoto et al., 2004); but, as we show below, the fluxes in debris disks are too low to detect features that are a few percent on the continuum, such as the phyllosilicates we consider here. We therefore do not consider debris disks further and turn our attention to protoplanetary disks. If protoplanetary disks contain predominantly interstellar dust, then their spectra would not yield any new information
968
MORRIS AND DESCH
a
b
c
d
FIG. 1. The absorption efficiency factor, Qabs, for (a) saponite, (b) serpentine, (c) montmorillonite, and (d) cronstedtite calculated from n and k determined by Glotch et al. (2007).
about the processes in the disk or the composition of planetesimals in them. Protoplanetary disks do not, however, contain pure interstellar dust. The existence of crystalline silicates argues strongly for thermal processing of dust within them (e.g., Wooden et al., 2007), as the interstellar medium contains only (>99.8%) amorphous silicates (Kemper et al., 2004). Dust samples from comet 81P=Wild 2, which were returned as a part of the STARDUST mission, have been shown to have a solar isotopic composition (Brownlee et al., 2006; Stephan, 2008; Zolensky et al., 2008), which indicates processing in the early Solar System. Observations of Orion proplyds also indicate growth of the minimum grain size in the first few105 years (Throop et al., 2001). This is not interpreted to mean that grains took >105 years to collide. Quite the contrary, micron-sized grains coagulate and fragment on timescales
